Research On Fault Diagnosis Algorithm Of Redundant Strapdown Inertial Navigation System Based On Improved GLT

In modern society,navigation system is more and more widely used in the field of national defense and economy,so the reliability requirements of navigation system are higher and higher.Fault diagnosis of Strapdown Inertial navigation system(SINS)with redundant configuration is an effective measure to improve the reliability of navigation system.The purpose of this paper is to study the solutions to the problems in the application of GLT in the fault diagnosis of redundant SINS.This paper focuses on the influence of scale coefficient error,installation error and constant bias of inertial sensor output on soft fault diagnosis when the inertial sensor of redundant SINS has soft fault,and verifies the effectiveness of the method of soft fault detection and isolation by compensating the odd and even vectors with BP neural network.According to the selected redundancy configuration scheme,the BP neural network model is determined,and the trained BP neural network is used to compensate the even and odd vectors.The simulation results show that the influence of sensor error on soft fault diagnosis is reduced by the compensation of even and odd vectors,and the soft fault which is difficult to detect without compensation is detected by the fault diagnosis system.Aiming at the problem that the statistical threshold adopted by GLT method can not adapt to the influence of environment and noise,which makes the false alarm rate and missed detection rate too large,and the reliability of inertial navigation system is reduced,this paper studies and proposes a method to select the fault detection threshold based on the cost function.By choosing different cost functions,the threshold selection scheme of the system under different working conditions is determined.By designing two cost functions,the minimum error probability method and the system maximum task reliability method,combined with the actual situation of the navigation task,when there are only four redundant sensors working normally,the threshold value that makes the system work at the minimum error probability is selected;when the number of redundant sensors working normally is more than four,the threshold value that makes the system work at the maximum reliability is selected.The simulation results show that the influence of environment and noise on the reliability of Redundant SINS can be reduced by using different thresholds in different situations.Considering the situation that multiple inertial sensors fail at the same time at a certain time,aiming at the problem that the traditional GLT algorithm can only detect and isolate single fault,but not deal with multiple faults,this paper proposes a method based on the total error vector,by improving the GLT fault isolation algorithm,so that the GLT fault diagnosis algorithm can determine the fault sensor number after detecting the fault.At the same time,different fault search and isolation processes are designed for different fault information of sensor.Simulation results show that GLT can detect and isolate multiple faults correctly.In this paper,by solving the problems in the application of GLT algorithm in the fault diagnosis of redundant SINS,the system has stronger adaptability to the environment and higher reliability.The research of this paper has some reference significance for the research of GLT algorithm and the application of redundant SINS.